Full Network Model Expansion Pre-implementation Analysis

Transcription

1 September 11, 2014

2 1. Executive summary As a condition of approving the initiative, the ISO Board of Governors requested a pre-implementation analysis that demonstrates the ISO s ability to model in the ISO market unscheduled flow from balancing areas outside the ISO. This document provides the analysis results which show that the ISO s modeling is an improvement over its existing practice of not modeling this unscheduled flow. The Federal Energy Regulatory Commission also requires that the ISO provide this analysis to it before implementing the market changes resulting from the initiative. 2. Background In February 2014, the ISO Board of Governors approved the initiative, which FERC subsequently approved on July 31, This initiative includes the following market enhancements: 1. Expanding the model of the physical electric network used by the ISO market to include the other balancing areas in the Western Electricity Coordinating Council area. 2. Modeling in the ISO market forecasted unscheduled power flows within the ISO balancing area based on an expanded network topology and caused by the load, generation, and interchanges forecasted for other balancing areas in the western interconnection. 3. Modeling unscheduled flow to produce feasible ISO market schedules and incorporating the unscheduled flow into ISO market prices. This includes incorporating physical flow limits over certain ISO interties into the ISO markets, where currently the ISO markets only enforce limits on scheduled flow. The modeling enhancements improve reliability because expanding the full network model will allow the ISO to more accurately model expected real-time conditions in the day-ahead timeframe by including unscheduled loop flow, outages, and contingencies. This aligns with Federal Energy Regulatory Commission and North American Electric Reliability Corporation recommendations after the September 8, 2011 southwest blackout that stated the ISO and other balancing areas should better coordinate their day-ahead planning. The market changes will also provide more accurate pricing by incorporating congestion caused by unscheduled loop flow and respecting the physical limits of the ISO s interties in the dayahead market. They will reduce infeasible schedules in the day-ahead market that result in redispatch of resources in the real-time market that contribute to uplift costs. The modeling of the external network also supports the feasibility of energy imbalance market schedules. 1

3 3. Purpose As part of the Board s approval of the Full Network Model initiative, the ISO committed to present a pre-implementation analysis so the Board can assess the accuracy of the ISO s modeling of unscheduled flow. The pre-implementation analysis will be presented at the September 2014 Board of Governors meeting. Specifically, the pre-implementation documentation states: The goal of this activity is to show that the calculated unscheduled flows provide a reasonable estimate for the actual unscheduled flows that materialize in real time and [is] ignored in the existing day-ahead market solution. 1 The Board also authorized the ISO to file any tariff changes with the Federal Energy Regulatory Commission to implement the requested modeling enhancements. The ISO did so and on July 31, 2014, the Commission conditionally approved the ISO s tariff amendments. 2 The Commission found the ISO s proposal just and reasonable but conditioned its acceptance on a compliance filing. The compliance filing includes the requirement to submit information on the calculation of an accuracy metric. 3 The accuracy metric compares the modeled versus actual unscheduled flow due to the modeling of other balancing areas in the Western Electricity Coordinating Council area in the ISO s expanded full network model. The accuracy metric assesses the accuracy of the ISO s modeling of external unscheduled flow over the ISO interties because of non-iso balancing area load, generation, and interchanges (imports and exports). The metric will be calculated for the ISO as a whole and will determine whether the ISO continues to model external unscheduled flow in its day-ahead market. Though the accuracy metric specified in the compliance filing only applies post-implementation, it captures the same intent as the ISO s proposed pre-implementation analysis. Given the Federal Energy Regulatory Commission s required compliance filing, the ISO can more efficiently address both the pre-implementation analysis and post-implementation metric by aligning the two processes. This allows the Board, stakeholders, and the ISO to preview the results of the post-implementation accuracy metric during the pre-implementation phase. Section 4 describes the methodology used to calculate the accuracy metric for the preimplementation analysis and Section 5 discusses the results. Section provides concluding thoughts and next steps. The Appendix includes detailed confidential data and is available subject to signing the WECC Universal Non-Disclosure Agreement. 1 California ISO,, Addendum:, January 23, 2014, pg Ca. Indep. Sys. Operator Corp. 148 FERC 61,089 (2014), Docket No. ER Other compliance filing requirements include: a triggering mechanism based on the accuracy metric which will require that the ISO not enforce physical flows on its interties in the day-ahead market, filing with the Commission the ISO s pre-implementation analysis provided to the Board prior to implementation, and minor tariff revisions agreed to by the ISO. 2

4 4. Methodology The accuracy metric compares the ISO s day-ahead modeling of external unscheduled flows to the actual external unscheduled flow under two scenarios: Scenario 1: The ISO models external unscheduled flow impacts in the day-ahead Scenario 2: The ISO does not model external unscheduled flow impacts in the dayahead For the purpose of this analysis, the ISO calculates day-ahead external unscheduled flow based on a power flow run that uses the same inputs as the solved day-ahead case as shown in Figure 1 below. 4 These inputs are calculated per external balancing area and include: the forecast net interchange (arrow 1 in Figure 1), demand forecast from the Reliability Coordinator (arrow 2), derived generation from the first two terms (arrow 3), and generation and load distribution factors (arrow 4). By using a solved day-ahead case, the ISO will also have the dayahead expanded full network topology (arrow 5) and planned outages (arrow 6) for the external balancing authorities and the ISO. The ISO then runs a power flow with these hourly inputs but excludes the ISO load, generation, and interchange. The output is the day-ahead external unscheduled flow impact per ISO intertie, per hour (DaExUSF tie,hour ). 4 Inputs for the day-ahead case are described in the draft final proposal and addendum to the draft final proposal available at: 3

5 Figure 1 Day-ahead external unscheduled flow calculation forecasted net scheduled interchange (no ISO schedules) demand forecast (no ISO forecast) External BAs derived generation (no ISO forecast) External BAs generation and load distribution factors DA FNM topology (Western BAs including ISO) Planned outages (Western BAs including ISO) Power flow (no ISO load, no ISO generation, no ISO interchange) DA external unscheduled flows per ISO intertie, per hour (DaExUSF tie,hour ) The ISO follows a similar process for the actual external unscheduled flow. Actual information is derived from the State Estimator and averaged to an hourly number to compare with hourly day-ahead data. The inputs are calculated per external balancing area and include: the average actual net interchange (arrow 1 in Figure 2), average actual demand (arrow 2), average actual generation (arrow 3), and generation and load distribution factors (arrow 4). By using a State Estimator solution, the ISO will also have the expanded topology and outages (arrow 5) for the external balancing authorities including the ISO. The ISO then runs a power flow with these inputs excluding the ISO load, generation, and interchange. The output is the average actual external unscheduled flow impact per ISO intertie, per hour (ActExUSF tie,hour ) as shown in Figure 2 below. 4

6 Figure 2 Average actual external unscheduled flow calculation average actual net scheduled interchange (no ISO schedules) average actual load (no ISO load) average actual generation (no ISO generation) generation and load distribution factors State Estimator FNM topology (Western BAs including ISO) Power flow (no ISO load, no ISO generation, no ISO interchange) Average actual external unscheduled flows per ISO intertie per hour (ActExUSF tie,hour ) The outputs from the processes in Figure 1 and Figure 2 are evaluated under two scenarios: (1) the ISO models external unscheduled flow impacts in the day-ahead; and (2) the ISO does not model external unscheduled flow impacts in the day-ahead. Table 1 below shows the variables needed to calculate the accuracy metric. 5

7 Table 1 Variables for and calculation of the accuracy metric Scenario Day-ahead Actual Calculation 1. ISO models external unscheduled flow impacts in the day-ahead 2. ISO does not model external unscheduled flow impacts in the day-ahead Day-ahead external unscheduled flows per ISO intertie, per hour (DaExUSF tie,hour ) Flow is zero (DaExUSF tie,hour = 0) Average actual external unscheduled flows per ISO intertie, per hour (ActExUSF tie,hour ) Average actual external unscheduled flows per ISO intertie, per hour (ActExUSF tie,hour ) Sum of the absolute value of the difference between day-ahead and actual external unscheduled flows per ISO intertie, per hour Scenario 1: DaExUSF tie,hour - ActExUSF tie, hour Scenario 2: 0 - ActExUSF tie,hour Sum of Scenario 1 is less than the sum of Scenario 2 over a 3-week rolling average PASS DaExUSF tie,hour - ActExUSF tie,hour < 0 - ActExUSF tie,hour PASS Under Scenario 1, the variables are the day-ahead and average actual external unscheduled flow outputs shown in Figure 1 and Figure 2, respectively. The ISO calculates the sum of the absolute value difference between these two variables per ISO intertie, per hour. Under Scenario 2, the ISO does not model external unscheduled flow impacts. Therefore, the flow is zero in the day-ahead. The average actual external unscheduled flow under Scenario 2 is the same as Scenario 1 per ISO intertie, per hour. The calculation is the same. These two calculations were made for every hour and every day of the pre-implementation analysis period. 5 The ISO passes the accuracy metric if the sum under Scenario 1 is less than the sum under Scenario 2 over a three-week rolling average. 5. Results In the pre-implementation analysis description from the addendum to the draft final proposal, the ISO proposed to analyze four interties: (1) California-Oregon Intertie; (2) Palo Verde; (3) Eldorado-Mead; and (4) Victorville-Lugo. The ISO also proposed to compare results for representative internal constraints. The original proposal was limited to four interties to balance 5 The accuracy metric makes an exception for hours in which there are extraordinary unforeseen real-time events such as: the loss of direct current transmission lines, unexpected outages of generators over 1,000 MW, or a derate of over 1,000 MW at any intertie. These hours may be removed from the analysis. 6

8 the work load during the pre-implementation and market simulation period. However, the ISO created an efficient process to calculate the accuracy metric and produced results for all ISO interties and selected internal constraints. Consequently, they are included in the analysis presented in this report. The ISO conducted the pre-implementation analysis using 12 operating days in August and September. This is later than the original dates proposed in the addendum because of delays in the market simulation period and the new obligation to conduct the post-implementation accuracy metric. No hours were removed from the analysis due to extraordinary real-time events. The operating days analyzed are not a continuous series of days for two reasons. Two days are not included because of spurious results produced by duplicate records in input data. The ISO modified its procedures to prevent this in the future. The ISO could not rerun the analysis for these days because of the timing of data available in its simulation systems. The other days not listed are not included because of the timing of data available in the ISO s simulation system. Table 2 below shows results for all ISO interties for each day of the analysis period. Columns [B] and [C] are daily external unscheduled flow impacts for Scenario 1 and 2, respectively. Columns [D] through [F] are the cumulative hours and external unscheduled flow impacts for each of the 12 days of analysis. Columns [G] and [H] calculate the rolling average by dividing the cumulative external unscheduled flow by the cumulative hours. Over this three week period, the ISO passed the accuracy metric because the final row in column [G] is less than the final row in column [H]. The data shown is aggregated for the entire ISO. More granular data is available in Appendix A subject to signing the WECC Universal Non-Disclosure Agreement. 7

9 Table 2 ISO intertie accuracy metric for 12 analysis days Daily Cumulative Rolling average Scenario 1 Scenario 2 Scenario 1 Scenario 2 Scenario 1 Scenario 2 Date Hours MWh MWh MWh MWh MW MW [A] [B] [C] [D] [E] [F] [G] [H] = [E] / [D] = [F] / [D] 8/16/ , , , ,398 2,101 4,475 8/19/ ,659 97, , ,553 2,044 4,262 8/20/ , , , ,221 2,211 4,239 8/24/ ,656 78, , ,838 2,186 3,998 8/25/ ,037 81, , ,049 2,149 3,875 8/26/ ,946 84, , ,517 2,138 3,816 8/27/ ,320 89, , ,144 2,150 3,804 8/28/ ,340 76, , ,056 2,096 3,729 8/29/ ,240 72, , ,927 2,054 3,652 8/30/ ,304 66, , ,358 1,908 3,564 8/31/ ,166 64, , ,900 1,796 3,484 9/3/ ,882 75, , ,218 1,688 3,456 Table 3 below shows results for selected ISO internal constraints for each day of the analysis period. Columns [B] and [C] are daily external unscheduled flow impacts for Scenario 1 and 2, respectively. Columns [D] through [F] are the cumulative hours and external unscheduled flow impacts for each of the 12 days of analysis. Columns [G] and [H] calculate the rolling average by dividing the cumulative external unscheduled flow by the cumulative hours. Over this three week period, the ISO passed the accuracy metric because the final row in column [G] is less than the final row in column [H]. The data shown is aggregated. More granular data is available in Appendix A subject to signing the WECC Universal Non-Disclosure Agreement. 8

10 Table 3 ISO selected internal constraint accuracy metric for 12 analysis days Daily Cumulative Rolling average Scenario 1 Scenario 2 Scenario 1 Scenario 2 Scenario 1 Scenario 2 Date Hours MWh MWh MWh MWh MW MW [A] [B] [C] [D] [E] [F] [G] [H] = [E] / [D] = [F] / [D] 8/16/ ,269 24, ,269 24, ,006 8/19/ ,492 30, ,760 54, ,130 8/20/ ,073 21, ,833 75, ,047 8/24/ ,300 24, ,133 99, ,042 8/25/ ,518 19, , , /26/2014 8,901 22, , , /27/2014 7,023 22, , , /28/ ,899 39, , , ,061 8/29/ ,761 36, , , ,114 8/30/2014 1,949 31, , , ,135 8/31/2014 7,709 30, , , ,147 9/3/2014 6,128 38, , , , Conclusion and next steps The ISO performed consistently well on the overall accuracy metric for the days analyzed. The results show that the ISO s proposed methodology is sound and that it will help the ISO reasonably estimate in the day-ahead the external unscheduled flow that will materialize in the real-time. This is an improvement over today s practice which does not model these flows. The analysis is also an improvement over the previously described methodology because it produced results for every intertie in addition to selected internal constraints, and provides insight into the ISO s projected performance under the post-implementation accuracy metric before the obligation starts. The ISO will hold a conference call with stakeholders on September 17, before the Board of Governor s meeting scheduled for September 18-19, 2014 to review the results. Stakeholders that are signatories to the WECC Universal Non-Disclosure Agreement may request the confidential appendix with detailed hourly and tie results by submitting a request to caisonda@caiso.com, listing their entity name as listed on the WECC Universal Non-Disclosure Agreement. At the Board of Governor s meeting, the ISO will present the results of this analysis in a briefing to Board members. The ISO will also file this report with the Federal Energy Regulatory Commission as part of its compliance obligation. 9